In the interest of maintaining a good supply voltage on the AREF pin, I am considering a voltage reference, the MAX6107. This little SOT23 + a 1uF capacitor is allegedly all I need to maintain a stable 4.5V. For those who have been through this before me, I wonder if I have this right...
On page 343 of the 328P datasheet (http://www.atmel.com/dyn/resources/prod_documents/doc8025.pdf), the AREF current requirement is listed as about 130uA @ 4.5V. The Maxim 6107 (http://datasheets.maxim-ic.com/en/ds/MAX6100-MAX6107.pdf) is allegedly good at putting out up to 5ma if I am reading that datasheet correctly. So, assuming we're driving nothing more than the 328P and two BAV99s off of the AREF source, the Max6107 should be OK, the AREF=4.5V, etc.
Or am I missing something?
I think the ATmega comes up with the ARef line connected to the Vcc line. This might cause problems with the other users of the analog reference voltage until you get a chance to switch the ARef to External. I think they recommend having the ARef voltage supplied thorough a resistor so while it is shorted to Vcc it doesn't draw a lot of current.
Thank you, John. What size resistor have you used? I cannot find a recommendation in the datasheet, but for all I know there is a specific application note that I may have missed.
In the past, I have used a LD39015M125R voltage regulator http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00173477.pdf providing 1.25V to AREF, a 1uF cap, and the two BAV99s protecting the inputs - and it worked. However, as I understand it a voltage reference is a different animal and hence may not survive what this regulator apparently has. Maybe I should just design around a voltage that the LD39015 series can provide?
For example, the LD39015M25R puts out 2.5V and I could just set up a voltage divider to take care of the rest...